Introducing RIDArena
by Sev, Leo Mckee-Reid, and Tyler Tracy
RIDArena (Rogue Internal Deployment Arena) is an AI control evaluation environment that simulates an AI lab’s R&D infrastructure. Researchers can use it to study how an agent can covertly sabotage the lab’s monitoring system and deploy another rogue agent without oversight. RIDArena can support sabotage, control, and monitor evaluations in...
Nov 145